Panel Unit-root Tests for Heteroskedastic Panels

Abstract

In this article, we describe the command xtpurt, which implements the heteroskedasticity-robust panel unit-root tests suggested in Herwartz and Siedenburg (2008, Computational Statistics and Data Analysis 53: 137–150), Demetrescu and Hanck (2012a, Economics Letters 117: 10–13), and, recently, Herwartz, Maxand, and Walle (2017, Center for European, Governance and Economic Development Research Discussion Papers 314). While the former two tests are robust to time-varying volatility when the data contain only an intercept, the latter test is unique because it is asymptotically pivotal for trending heteroskedastic panels. Moreover, xtpurt incorporates lag-order selection, prewhitening, and detrending procedures to account for serial correlation and trending data.

Threshold effect of non-interest income disaggregates on commercial banks’ financial performance in Zimbabwe

“…Table 4 presents Clemente et al [ 120 ], Levin et al [ 121 ], and Herwartz et al [ 122 , 123 ] panel unit root test results. The nature of our data demands a battery of robust panel unit root tests since it is plagued with traits of heteroscedasticity and structural breaks possibly emanating from shocks triggered by currency and economic transitions over the sample period.…”

Testing the validity of Gibrat’s law for Slovenian farms: cross-sectional dependence and unit root tests

“…Thus, first, we employ second generation panel unit root tests to check the unconditional forms of Gibrat's law in Slovenian farm growth. In addition, we apply different heteroskedasticity-robust panel unit-root tests suggested in Herwartz and Siedenburg (2008), Demetrescu andHanck (2012), andHerwartz, Maxand, Raters, andWalle (2018).…”

The Effectiveness of Climate Adaptation Finance and Readiness on Vulnerability in African Economies

“…The test yields more reliable results when dealing with disturbance processes that may exhibit heteroscedasticity and autocorrelation in our dataset [103,104]. Most panel unit root tests, including the Breitung and Das test and Levin-Lin-Chu unit root test, assume homoskedasticity [105] and hence were not applicable.…”